Most agent tutorials end at a notebook cell. The agent answers one question, in an environment only its author can reach, and then the laptop closes and it stops existing. What is missing sits around the model rather than inside it: a memory that survives the session, a schedule that fires with nobody present, and a channel that reaches you when you are nowhere near a screen.
The first evening is the groundwork: the building blocks of agentic systems, the six patterns we keep returning to, and the case studies behind them, drawn from the same material as our published work. The second evening is the code-along, where we build the terminal over live IDX and SGX data and put it somewhere it stays up, which is the part most courses leave out. By the end it holds a conversation on Telegram and sends a briefing before the market opens, tailored to your personal analysis framework and watchlist.
Building Blocks and Agentic Patterns
The established paradigms behind robust Agentic AI systems.
- The building blocks every Agentic AI system is assembled from.
- Six agentic patterns, and the conditions that make each one right.
- Worked examples from enterprise systems we have put into production.
The Personal Finance Terminal Blueprint
How those patterns compose into an agent that watches a market.
- Tools, memory, and retrieval arranged around live market data.
- Guardrails for a domain where a confident wrong answer is costly.
- What has to hold true before an agent can be left running unattended.
Engineering Practices and Code-Along
Building the terminal together, and leaving it running.
- Writing the agent loop, the tool schemas, and the guardrails together.
- Wiring in live IDX and SGX data as tools the agent can reach for.
- Connecting Telegram, so that the agent has an address of its own.
- A watchlist, along with conditions that decide when to send message.
- Deploying it so that the briefing arrives without manual intervention.
🍀 Learning Objectives
- Established Agentic Patterns
The building blocks and the six recurring patterns behind robust Agentic AI systems, drawn from our publications and the systems we run for clients.
- Agents Powered by Live Market Data
Give an agent tools that read live IDX and SGX figures, so that its answers rest on numbers anyone can check.
- Your Own Personal Finance Terminal
Build and deploy an agent that holds a conversation on Telegram and sends a briefing before the market opens.



